An online sketching and gesture editing system for conceptual design

The paper describes the development of an online sketching design and gesture editing system for conceptual design, named Sketch2D. The system uses sketches and gestures as the main interaction paradigm. First, the preprocessing methods of the single sketching stroke are presented including automatic segmentation, recognition and regularity. Then we design nine kinds of pen-based gestures which are used to edit the conceptual sketch. At last, the experimental results of three different training algorithms for gesture recognition are compared with each other such as pattern matching, Bayes classifier, BP neural network. Experiments show that the presented online sketching system can improve user adaptation and intelligence of current CAD systems